Hi, some time ago I've reported a problem with dag_shortest_paths that occurs when I'm searching for the *longest* path by appropriately changing inf/compare operators. I've fixed that problem by change in dag_shortest_paths.hpp, but looks like it's back after changes to relax.hpp. I attach my original report, which has a test program. When using version 1.21 of graph/relax.hpp, it produces the expected outcome '2'. When using version 1.22, I get 2147483647 Revision 1.22 is: revision 1.22 date: 27/10/2004 18:38; author: dgregor Merge from graph_devel_1_33_0 branch And the only commit on that branch does not clarify anything either: revision 1.21.8.1 date: 15/10/2004 18:54; author: dgregor Added several algorithms: - random graph layout - Fruchterman-Reingold layout - A* search - Floyd-Warshall all-pairs shortest paths Fixes for several algorithms: - Documentation for Johnson all-pairs shortest paths - Bellman-Ford single-source shortest paths Any ideas? - Volodya
On Jun 17, 2005, at 8:59 AM, Vladimir Prus wrote:
some time ago I've reported a problem with dag_shortest_paths that occurs when I'm searching for the *longest* path by appropriately changing inf/compare operators.
I've fixed that problem by change in dag_shortest_paths.hpp, but looks like it's back after changes to relax.hpp.
Oh, darn. Thanks for reporting it (again).
I attach my original report, which has a test program. When using version 1.21 of graph/relax.hpp, it produces the expected outcome '2'. When using version 1.22, I get 2147483647
I've added your test program to the regression tests and rolled back relax.hpp (fixing a few other algorithms in the process). Now it shouldn't happen again :) Doug
participants (2)
-
Doug Gregor -
Vladimir Prus